www.afm-journal.de
P. T. P. Le, K. Hofhuis, Dr. A. Rana, Prof. M. Huijben,
Prof. H. Hilgenkamp, Prof. G. A. J. H. M. Rijnders, Prof. J. E. ten Elshof, Prof. G. Koster
MESA+ Institute for Nanotechnology University of Twente
PO Box 217, 7522 NH Enschede, The Netherlands E-mail: G.Koster@utwente.nl
Dr. N. Gauquelin, G. Lumbeeck
Electron Microscopy for Materials Science (EMAT) University of Antwerp
2020 Antwerp, Belgium
E-mail: Nicolas.Gauquelin@uantwerpen.be Dr. C. Schüßler-Langeheine
Helmholtz-Zentrum Berlin für Materialien und Energie BESSY II
Albert-Einstein-Str. 15, 12489 Berlin, Germany
Vanadium dioxide (VO2) is a much-discussed material for oxide electronics and neuromorphic computing
applications. Here, heteroepitaxy of VO2 is realized on top of oxide nanosheets that cover either the amorphous
silicon dioxide surfaces of Si substrates or X-ray transparent silicon nitride membranes. The out-of-plane orientation of the VO2 thin films is controlled at will between (011)M1/(110)R and (−402)M1/(002)R by coating the bulk substrates
with Ti0.87O2 and NbWO6 nanosheets, respectively, prior to VO2 growth. Temperature-dependent X-ray diffraction
and automated crystal orientation mapping in microprobe transmission electron microscope mode (ACOM-TEM) characterize the high phase purity, the crystallographic and orientational properties of the VO2 films. Transport
measurements and soft X-ray absorption in transmission are used to probe the VO2 metal–insulator transition,
showing results of a quality equal to those from epitaxial films on bulk single-crystal substrates. Successful local manipulation of two different VO2 orientations on a single substrate is demonstrated using VO2 grown on
lithographically patterned lines of Ti0.87O2 and NbWO6 nanosheets investigated by electron backscatter diffraction.
Finally, the excellent suitability of these nanosheet-templated VO2 films for advanced lensless imaging of the metal–
insulator transition using coherent soft X-rays is discussed.
Tailoring Vanadium Dioxide Film Orientation Using
Nanosheets: a Combined Microscopy, Diffraction,
Transport, and Soft X-Ray in Transmission Study
Phu Tran Phong Le, Kevin Hofhuis, Abhimanyu Rana, Mark Huijben, Hans Hilgenkamp,
Guus A.J.H.M. Rijnders, Johan E. ten Elshof, Gertjan Koster,* Nicolas Gauquelin,*
Gunnar Lumbeeck, Christian Schüßler-Langeheine, Horia Popescu, Franck Fortuna,
Steef Smit, Xanthe H. Verbeek, Georgios Araizi-Kanoutas, Shrawan Mishra,
Igor Vaskivskyi, Hermann A. Dürr, and Mark S. Golden*
DOI: 10.1002/adfm.201900028
The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adfm.201900028.
Dr. H. Popescu Synchrotron SOLEIL
L’Orme des Merisiers Saint-Aubin BP 48 91192 Gif-sur-Yvette Cedex, France F. Fortuna
CSNSM
Université Paris-Sud and CNRS/IN2P3
Bâtiments 104 et 108, 91405 Orsay Cedex, France S. Smit, X. H. Verbeek, G. Araizi-Kanoutas, Dr. S. Mishra, Prof. H. A. Dürr, Prof. M. S. Golden
Van der Waals-Zeeman Institute Institute of Physics
Science Park 904, 1098 XH Amsterdam, The Netherlands E-mail: M.S.Golden@uva.nl
Dr. S. Mishra
School of Material Science and Technology Indian Institute of Technology (BHU) Varanasi 221005, India
Dr. I. Vaskivskyi, Prof. H. A. Dürr Department of Physics and Astronomy Uppsala University
Box 516 751 20 Uppsala, Sweden Dr. I. Vaskivskyi
Center for Memory and Recording Research University of California San Diego
9500 Gilman Drive, La Jolla, CA 92093-0401, USA © 2019 The Authors. Published by WILEY-VCH Verlag GmbH & Co.
KGaA, Weinheim. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.
www.afm-journal.de www.advancedsciencenews.com
1. Introduction
Vanadium dioxide (VO2) has been drawing attention since the
discovery of its metal–insulator transition (MIT), signaled by a several orders of magnitude resistivity change close to 340 K.[1,2]
Given these remarkable properties, it may not be a surprise that VO2 is a leading candidate material for the development
of oxide devices for both low power electronics (high off-resist-ance), either in a more conventional field effect type of device or alternatively, neuromorphic electronic architectures[3,4] as a
memristive material. It has been known that the MIT occurs alongside an abrupt, first-order structural phase transformation from a metallic, tetragonal rutile (R) phase (P42/mnm) to an
insulating, monoclinic (M1) phase (P21/C). Recent work[5]
points out that this transition is preceded by a purely electronic softening of Coulomb correlations within the V–V singlet dimers that characterize the insulating state, setting the energy scale for driving the near-room-temperature insulator–metal transition in this paradigm complex, correlated oxide.
Up to now, most studies of epitaxial VO2 thin films have
used Al2O3 and TiO2 single crystal substrates to control film
orientation,[6,7] bringing along challenges of cost, limited size,
and incompatibility with the current Si-based technology for future VO2-based devices. Direct deposition of VO2 on glass
or Si substrates with a native amorphous silicon dioxide layer leads to a polycrystalline film with predominant (011)M1
orien-tation,[8] whereas VO
2 is favorably grown (010)M1-oriented when
a buffer layer of Pt(111) is used on a Si substrate.[9] Epitaxial
growth of VO2 with (010)M1 orientation is possible on epitaxial
layers of yttria-stabilized zirconia—YSZ(001)—on Si(001).[10,11]
Ideally, one would wish for direct control over VO2 film
ori-entation on Si or even on arbitrary substrates at will, without any concessions being made on the VO2 film quality. Oriented
growth is not only an important enabler for the fundamental study of the mechanism of the VO2 MIT,[12] but also for
poten-tial applications for next-generation transistors,[13] memory
metamaterials,[14] sensors,[15] and novel hydrogen storage
tech-nology.[16] Recently, various metal oxide films have been
suc-cessfully grown on glass and Si substrates using epitaxy on the so-called oxide nanosheets.[17–20] Oxide nanosheets are essentially
2D single crystals with a thickness of a few nanometers or less, and lateral size in the micrometer range. They can be made spanning a wide range of crystal lattices and 2D structural sym-metries,[21] allowing for new possibilities to tailor the important
structural parameters and properties of thin films on arbi-trary—and thus also technologically relevant—bulk substrates. Successful implementation of nanosheets in fact means that the choice of the bulk substrate becomes a free parameter that can enter the engineering cycle of each individual application.
In the research presented here, Ti0.87O2 (TO) and NbWO6
(NWO) nanosheets have been identified as being ideal tem-plates for the orientation of thin films of the important complex oxide VO2 on varying substrates. Monolayers of nanosheets
were deposited on Si substrates and alternatively on 20 nm thick Si3N4 transmission electron microscope (TEM) grids
using the Langmuir–Blodgett (LB) method. Then, utilizing pulsed laser deposition (PLD), single-phase VO2 thin films
were grown epitaxially on both TO and NWO nanosheets with (011)M1 [(110)R] and (−402)M1 [(002)R] out-of-plane orientation of
the low temperature monoclinic M1 phase [high-T rutile phase], respectively. The high structural and orientational quality of the VO2 made possible by the nanosheet epitaxy was proven using
TEM and X-ray diffraction across the Mott MIT, as well as by electron backscatter diffraction (EBSD) studies. In addition, both transport and soft X-ray spectroscopic probes of the MIT showed data of excellent quality, matching those for VO2 grown
on bulk single-crystalline substrates. Importantly, the use of a nanosheet-coated Si3N4 membrane as a PLD substrate allowed
soft X-ray absorption spectroscopy (XAS) to be carried out in the fully bulk-sensitive and highly direct transmission mode. Finally, the nanosheet-approach is shown to provide a high degree of control of the crystallographic orientation of the VO2
film. This is illustrated by the use of a single film-growth run to generate two different orientations on a single substrate deter-ministically, by arranging both TO and NWO nanosheets in an alternating, stripe-like pattern using lithography.
2. Results and Discussion
Figure 1 shows a plan view of the TO and NWO nanosheet
planes in panels (a) and (d), respectively. For the TO(NWO) nanosheets, the relevant 2D unit cell is highlighted in green (pink). In panels (b) and (c), the relevant VO2 planes are shown
for the M1 and R structures with the 2D unit cell of the TO nanosheets superimposed. Likewise, in panels (e) and (f) the 2D unit cell of the NWO nanosheets is superimposed on the relevant planes in the M1 and R VO2 phases. What these
fig-ures show, backed up by the data of Table 1, is that the (011)M1
[(110)R] and (−402)M1 [(002)R] out-of-plane orientations of M1[R]
VO2 are compatible with the TO and NWO nanosheet
sym-metry and lattice constants, respectively.
In the following, the reasoning and data that led to this con-clusion are gone through in a step-by-step manner. A closer examination of Figure 1 shows that—considering the oxygen sub-lattices as the dominating structural entities driving potential epitaxy between the VO2 and the nanosheets—the
two distinct OO distances of 2.90 and 2.84 Å for the VO2
(011)M1 plane are close to the OO distance of 2.97 Å for the
TO nanosheet in the b-direction. In the a-direction, there are also two distinct OO distances for the VO2 (011)M1 plane, i.e.,
3.38 and 4.51 Å, respectively, which are very different from the OO distance of 3.82 Å for TO nanosheets. Highly relevant in this regard is the concept of domain matching epitaxy,[22] which
can be seen as a generalization of the more common lattice match epitaxy. In domain matching epitaxy, integral multiples of lattice planes match across the film–substrate interface, with the size of the domain (equaling integral multiples of planar spacing), determined by the degree of mismatch. This can yield epitaxy when mismatches exceed the usual 7–8% limit for reg-ular lattice epitaxy.[22]
If we consider the domain epitaxy approach, it can be seen that the slightly distorted (only 0.35° away from a right angle) rectangular oxygen domain of 7.56 × 5.74 Å2 for the
VO2 (011)M1 plane is close to that of 7.64 × 5.94 Å2 for the TO
nanosheets. This results in a small domain mismatch of only 1% and 3.4% in the a- and b-directions at room temperature, respectively, suggesting that the M1 monoclinic phase of VO2
(011)M1 film can be stabilized by TO nanosheets at room
tem-perature. At the growth temperature of 520 °C, it is the rutile phase of VO2 that is being formed,[2] and analogous
consider-ations show that as the (110)R plane is very closely related to
the room temperature (011)M1 plane of VO2 a similar domain
epitaxial relationship will be at work.
In addition to the domain epitaxy ideas, we also point out that as the nanosheets are exfoliated layered materials without chemically active dangling bonds, there are only isotropic Cou-lomb and/or Van der Waals interactions between the growing film and the nanosheet surface. Consequently, lateral adatom– adatom interactions are significant drivers of the energetics of the early stages of (epitaxial) growth, thus helping to favor epitaxial growth also in the presence of relatively large lattice mismatch. In this manner, retention of epitaxy with lattice mis-match as high as 13% have been reported,[19,23] supporting the
ability of TO nanosheets to successfully template the (110)R
growth of the rutile phase of VO2.
We now turn to the NWO nanosheets. In this case, epi-taxial growth of (−402)M1 or (002)R VO2 films are expected via
straightforward lattice matching considerations. The 2D atomic structure of the (−402)M1 plane is distorted away from a square
planar symmetry (by 3° off the right angle). The lattice mis-match between this plane and the nanosheets is 3.6% and 7.9%
in the a- and b-directions, respectively, which—especially given the arguments above as regards the nondirectional nature of the adatom-nanosheet interactions—we would expect to sup-port epitaxial growth with the VO2 film oriented in the [−402]M1
direction at room temperature. For the NWO case, the high-T situation is simpler still, as the lattice mismatch between the high temperature (002)R plane and the NWO nanosheet is only
3.2%, with the same 2D square atomic structure present in both cases.
To summarize this section: detailed consideration of the oxygen sub-lattice-driven epitaxial relationships shows that for TO nanosheets, a combination of domain matching epitaxy and the weak adatom–nanosheet interactions should enable epitaxial growth of VO2, with the out-of-plane orientation being
(011)M1 and (110)R at room temperature and elevated
tempera-ture, respectively. For the NWO nanosheets, the VO2
orienta-tion is expected to be (−402)M1 and (002)R.
Turning to the first step of the nanosheet-templated film growth process in practice, Figure 2 shows atomic force microscopy (AFM) images of the morphology of the bare-nanosheets of TO and NWO in panels (a) and (b). It is clear that monolayers of TO and NWO nanosheets can be fabricated on Si substrates successfully with a surface coverage exceeding 95%. The lateral size of the individual nanosheets is 3–5 µm for both TO and NWO, which is partly governed by the grain size of the parent layered crystals obtained by solid state reaction.[24] The exfoliated nanosheets were deposited using a
LB-method to form a monolayer film on Si substrates and on Si3N4 membranes.
Using these nanosheet layers on Si3N4 TEM grids as
sub-strates, thin films of VO2 were grown using PLD. In order Figure 1. Excellent epitaxy conditions for nanosheets and VO2. Schematic illustration of atomic structures for a) TO nanosheet, b) VO2 (011)M1, and
c) VO2 (110)R. A suitable TO unit cell is shown shaded in green. d) NWO nanosheet, e) VO2 (−402)M1, and f) VO2 (002)R. A suitable NWO unit cell
is shown shaded in pink.
Table 1. Symmetry and lattice constants of TO and NWO nanosheets. Nanosheet 2D structure Lattice constant [Å] Ti0.87O20.52− Rectangular a = 3.76; b = 2.97
www.afm-journal.de www.advancedsciencenews.com
to determine the homogeneity and epitaxial quality of the VO2 films with nm lateral spatial precision, orientation maps
were recorded using TEM, the results of which are shown in Figure 2c–f. The data were measured in the rutile VO2 phase at
a temperature of 423K. For the TO-templated system, a remark-ably homogeneous (110)R orientation of the VO2 film results
(Figure 2c), and apart from a very few rogue patches, Figure 2d shows that the NWO nanosheets perform just as well, gener-ating (002)R VO2. These results are confirmed in the pole
fig-ures for these two rutile orientations shown in panels (e and f). With this nanoscopic confirmation of the excellent epitaxial growth on each individual nanosheet, the next step is a more global measure of the VO2 structure, in both the rutile and
monoclinic phases using X-ray diffraction (XRD).
Figure 3 presents the XRD data measured below and above TMIT of VO2 films grown using PLD at 520 °C on monolayers
of TO (panel (a)) and NWO nanosheets (panel(b)) on Si sub-strates. At 303 K, the peaks seen at 2θ values of 27.90° and 57.72° in Figure 3a are from the (011)M1 and (022)M1 VO2
reflections, respectively. In Figure 3b, the peak at 65.07° corre-sponds to the (−402)M1 VO2 reflection. At 403K—above the MIT
temperature—peaks measured at 27.70°, 57.27° and 65.28° now correspond to the (110)R, (220)R, and (002)R Bragg peaks
of rutile VO2, respectively. These XRD data confirm the
excel-lent orientational integrity of the VO2 films at the macroscopic
scale, confirming what was seen in the TEM data. This points toward the major role of the oxygen framework in the deter-mination of the epitaxial relationships between both nanosheet systems and the VO2 overlayer, as discussed earlier.
One of the motivations for choosing VO2 for this study
was its dual role as a model system for both understanding strongly correlated MIT, as well as its tunable/switchable large resistance change near room temperature.[3]
Conse-quently, the transport behavior of VO2 grown using PLD on
nanosheet templates is of great interest, and these data are shown in Figure 4. Defining the midpoint of the transition in the resistance curve measured upon heating as the phase transition temperature, TMIT, our data show transition
tem-peratures close to the canonical value of 341 K for bulk VO2
single crystals.[1,2] The T
MIT of the films with out-of-plane (110)R
texture (rutile c-axis in-plane) grown on TO nanosheets was 347 K, whereas for the out-of-plane (002)R textured film on
NWO nanosheets TMIT was 332 K. These values are in
agree-ment with what has been found in the literature on the orienta-tion dependence of TMIT on different bulk TiO2 substrates.[25]
The TMIT in VO2 is related to the VV distance along the
rutile c-axis, which affects the orbital overlap and the metallicity in the rutile phase.[26] In the past, this had been studied as a
function of film orientation,[25] chemical doping,[26] thickness[27]
and strain (001)R[28] for films grown on single crystal substrates.
For example, under compressive strain, the TMIT of a 24 nm
thick (001)R VO2 film grown on TiO2 decreased to 330 K, while
it was the same as the bulk VO2 value of 341 K for a fully relaxed
74 nm thick film.[27] When the rutile c-axis is under tensile strain,
similar to using TiO2 (110) single crystal substrates, TMIT is seen
to increase to 350–369 K.[25] Therefore, as T
MIT for VO2 grown
on TO nanosheets is higher than that of strain-free, bulk VO2,
this indicates that the R-VO2 c-axis (2.86 Å when unstrained) is
under a degree of tensile strain in the [010] direction of the TO nanosheet which has b = 2.97 Å. The situation is the other way around for the VO2 grown on NWO nanosheets: here TMIT is
lower, suggesting compressive strain along the rutile c-axis, as could be expected from the Poisson effect given the tensile strain along the [100] and [010] directions of VO2 (a = 4.53 Å) due to
coupling to the NWO nanosheet (a = 4.68 Å).
The MIT of the TO-templated VO2 shows a steep-sided
hyster-esis curve. In comparison, the data of Figure 4 show that the VO2
grown on NWO nanosheets displays a relatively broad MIT. A
Figure 2. AFM of nanosheets and automated crystal orientation mapping of VO2 using TEM. The top (bottom) row of figures is from samples with
TO(NWO)-templates. AFM images of the a) TO and b) NWO nanosheets showing the homogeneous coverage (scale bar: 5 µm). Panels (c) and (d) show high temperature automated crystal orientation mapping in microprobe TEM mode (ACOM-TEM) characterization of the rutile VO2 phase at 423K (scale bar:
200 nm). e and f) Corresponding pole figures for the (110)R and (002)R directions, respectively, with a color-scaled intensity as shown in the index on the right
possible explanation for this observation could be related to the fact that the NWO-templated VO2 possesses a surface roughness
of order 10 nm, compared to 2 nm for the TO-templated VO2
films (see Figure S1 (Supporting Information) for AFM data from the VO2 films). As the MIT in VO2 possesses a strongly
percola-tive character,[29] the increased roughness, on top of an increased
density of grain boundaries due to inter-nanosheet boundaries[17]
can be at least a partial explanation of the broader transition for the case of VO2 grown on NWO nanosheet templates. In
addi-tion, the larger deviation from 90° bond angles of the 2D atomic structure of the (−402)M1 plane in the NWO case compared to
the (011)M1 plane for TO-templated growth (see Figure 1) is also
compatible with a larger domain/grain boundary contribution to the transport for the NWO- templated VO2.
From the experimental data thus far, it is clear that the nanosheets provide an elegant and effective method to con-trol the orientation of VO2 films on two widely different bulk
substrates. In order to both emphasize the added possibilities that freedom from “hard” substrate epitaxy enables, as well as to illustrate how this can also be controlled on the micrometer scale, the next section reports a new strategy to manipulate dif-ferent orientations of VO2 on a single, arbitrary substrate. To
do this, NWO nanosheets were lithographically line-patterned on top of a monolayer of TO nanosheets, with subsequent VO2 growth on this structured template layer to demonstrate
the local manipulation of the orientation of the resulting high-quality VO2 film.
Figure 5 presents the high resolution scanning electron
microscopy (HR-SEM) cross-sectional view of the line-patterned VO2, showing in panel (a) that the film thickness was roughly
50 nm on both types of nanosheets. In addition, the HR-SEM plan-view (Figure 5b) clearly reveals the different surface mor-phologies alluded to earlier on either side of the nanosheet boundary, consistent with the AFM data of VO2 films grown on
“single species” TO and NWO nanosheets (see Figure S1, Sup-porting Information). In both situations the resulting VO2 film
has a smoother surface on TO than on NWO nanosheets. In addition to SEM imaging, EBSD maps were recorded, providing crystallographic information on the VO2 film on
the lithographically patterned nanosheet layers, shown in Figure 5c,d. As these data were recorded at room temperature, we use in the following the M1 film orientations (the corre-sponding R-phase orientation was shown in Figure 2). The inverse pole figure maps reveal the out-of-plane orientations (−402)M1 (shown in gold color, becoming (002)R at elevated
temperature) and (011)M1 (shown in purple, becoming (110)R
at elevated temperature). Taking a closer look at a domain on single-typed nanosheet regions, the four inverse pole figures
Figure 3. Temperature-dependent identification of the VO2 structural
phases using X-ray diffraction. XRD patterns of VO2 films on a) TO and
b) NWO nanosheets, measured at 303 K (M1 phase) and 403 K (R phase). The three peaks labeled “*” originate from the Inconel alloy 625/718 clamps holding the sample on the diffractometer heating stage. The VO2
Bragg peaks can be indexed using the film orientation discussed in the text, fully in agreement with the TEM data of Figure 2 and literature values of the temperature dependence of the VO2 crystal structure.
Figure 4. Transport characterization of the Mott metal–insulator
tran-sition of VO2. Resistance ratio (R[R phase]/R[M1 phase]) of VO2 films
grown on TO and NWO nanosheets as a function of temperature. For the M1 phase, the resistance was 2.6 × 105Ω (5.6 × 104Ω) for TO (NWO)
nanosheet-templated VO2, respectively. The resistance in the VO2-R
phase was a factor 810 (280) times lower than in the M1-phase for TO
www.afm-journal.de www.advancedsciencenews.com
of Figure 5e correspond to the four chosen regions indicated in Figure 5c. These show the presence of a single out-of-plane orientation for each type of nanosheet. It is evident that the VO2 film out-of-plane orientations follow the underlying
nanosheets with high fidelity, down to the micrometer level as we designed.
The in-plane EBSD orientation map displayed in Figure 5d shows a random orientational distribution, resulting from the randomness of in-plane orientation of nanosheets during LB deposition. On each single nanosheet, the reduced symmetry of the monoclinic VO2 lattice when cooling down from the
high-temperature rutile phase is expected to induce the formation of four in-plane structural domains.[30] The inverse pole figures
shown in Figure 5f show two of these because, due to sym-metry (presence of a mirror plane) of the monoclinic phase,
only half of the complete stereographic projection is displayed. Furthermore, in domain 2, only 1 orientation is seen as we are looking along the 21 axis.
Having proven the excellent crystalline quality and the exqui-site control over the orientation of VO2 grown on nanosheets
of different types, spectroscopy in the soft X-ray regime is now used to benchmark the epitaxial samples grown on TO nanosheets and provide comparison of the sample quality to what is known in the literature. As shown in the data of Figure 2c–f, the nanosheet approach enables deposition of high-quality VO2 on Si3N4 membranes that are soft X-ray
trans-parent, opening a route to conducting XAS in transmission. This yields a bulk-sensitive and direct measure of the absorp-tion that can directly be correlated with local measurements carried out in the TEM. The majority of previous XAS studies
Figure 5. Micrometer-level, deterministic control over the VO2 orientation on a single substrate. Room-temperature HRSEM images of the
a) cross-section and b) plan views at the line boundary of VO2 film (left-hand side is VO2 grown on NWO nanosheet; right-hand side is growth on
TO nanosheet). Panels (c) and (d) show inverse pole figure maps, measured using EBSD at room temperature, displaying the VO2 film orientation
in the c) out-of-plane and d) in-plane directions. In panel (c), the left-hand and right-hand sides of the image show growth on NWO nanosheet (gold color). Here, the film normal is the monoclinic (−402) axis (→(002)R). The central, purple-colored strip represents growth on TO nanosheet where
the film normal is (011)M1, equivalent to (110)R. Panel d) shows the in-plane texture of the VO2 films along the horizontal direction (x). The scale
bars represent 100 nm in (a), 500 nm in (b), and 10 µm in both (c) and (d). Panel (e) shows out-of-plane inverse pole figures corresponding to the four different regions marked on panel (c). The out-of-plane orientation is clearly controlled by the type of nanosheets used. Panel (f) displays in-plane inverse pole figures corresponding to the same four regions marked in panel (d). Four different domains on each individual nanosheet result from the reduced monoclinic symmetry compared to the tetragonal rutile case. Only half of the stereographic projection is shown due to symmetry resulting in only two spots.
of VO2 have used indirect methods such as Total Electron Yield
(TEY) to monitor the X-Ray absorption process.[5,31–35]
Linking to the transport data presented in Figure 4, XAS at the vanadium-L2,3 (2p→3d transitions) and oxygen-K (1s→2p
transitions) absorption edges also directly probe the MIT, and are readily accessible using soft X-rays provided by a synchro-tron light source. In correlated transition metal oxides such as VO2, the ability of soft X-ray spectroscopy to provide detailed
information on the manifold and coupled degrees of freedom (e.g., lattice, spin, charge and orbital) using the transition metal-L2,3 and O-K edges have been studied extensively.[5,30–35]
Panel (a) of Figure 6 shows V-L2,3 edge data both for the
metallic (rutile) and insulating (monoclinic) phases for the TO nanosheet templated VO2. As reported by Aetukuri et al.,[27] the
changes seen to occur across the MIT are related to the orbital occupation, and in particular they underscore the transforma-tion of the 3D rutile situatransforma-tion to one in which V-dimers form along the direction of the rutile c-axis, leading to shifts and splitting of electronic states related to the orbitals polarized in this direction, referred to as the d|| states. For grazing
inci-dence (E⊥crutile) the temperature dependent changes shown in
Figure 6b agree excellently with published data on films grown on single crystalline, bulk substrates,[27] attesting to the quality
of the nanosheet templated VO2 thin films. Closer examination
of the insets yields that the bulk (transmission) X-ray absorption spectroscopy (XAS) shows an onset of the MIT on warming at 340 K, and that ≈40 K of further heating are required to com-plete the conversion to the rutile phase. The transport data shown in Figure 4 from fully analogous VO2 films show an
earlier onset and faster completion of the transformation on heating. This can be understood straightforwardly as resulting from the percolative nature of the transport probe on the one hand and to the bulk-sensitive, volume-fraction-driven absorp-tion of soft X-ray radiaabsorp-tion on the other hand.
One of the clear order parameters for the MIT is the opening of an energy gap in the insulating phase. This can be clearly seen in XAS at the OK edge, as shown in Figures 6b and 7b,c, and reported in the literature recently.[5] With reference to
the electronic structure schematic shown under the data of Figure 6b, the most marked spectroscopic changes in the OK edge spectra while entering the metallic phase are due to the closure of the gap, and the disappearance of the unoccupied d||*
states, the latter present in the monoclinic phase due to a split-ting of the highly directional d|| states. The insets to Figure 6b
show how these changes to the gap [d||* states] leads to an
increase (decrease) of the XAS absorption at the characteristic energy of 529.1 (530.6) eV.
Figure 7a illustrates the experimental configuration used
for polarized XAS in transmission. For the TO-nanosheet-tem-plated VO2 films, the rutile c-axis is in the film plane, and its
in-plane orientation varies from one nanosheet to the next. The synchrotron X-ray beam is large enough to average over a large number of nanosheets, meaning that we can consider vertically polarized radiation (LV) at grazing incidence to yield an unpo-larized spectrum. Horizontally pounpo-larized radiation (LH) aligns the E-vector perpendicular to the film plane and hence E⊥crutile
is always realized, regardless of the direction of the in-plane ori-entation of the rutile c-axis in each individual nanosheet-tem-plated epitaxial grain. In Figure 7b,c, polarization-dependent XAS spectra are shown for both grazing and normal incidence of the beam, respectively. As is clear from the earlier discus-sion, for normal incidence (Figure 7c), whether LV or LH radia-tion is used makes no difference to the absorpradia-tion spectra, as in all cases a mixture of E||crutile and E⊥crutile is the result. For
grazing incidence (Figure 7b) and LH radiation, the E vector is E⊥crutile, compared to mixed E⊥crutile and E||crutile for linear
vertical. This is of high relevance for future experiments such as those outlined in Figure 8 in the next section. The major
Figure 6. Soft X-ray absorption in transmission. V-L2,3 XAS of nanosheet-supported VO2 recorded in transmission at the temperatures shown for
grazing incidence of the linearly polarized X-rays. Subtle yet clear differences in the spectral features mirror alterations in the electronic structure, reflecting changes in orbital energies as the V-d|| states split and an energy gap opens in the insulating, low temperature phase. The details of the
differ-ence spectrum agree very well with published polarization-dependent V-L2,3 X-ray absorption data from VO2 grown epitaxially on bulk single crystalline
substrates.[27] Panel (b) shows the O-K edge recorded at normal incidence as a function of temperature. The insets—whose data points are color-coded
to match the spectra from which they are taken—show the T-dependence of the two main absorption features signaling the MIT. Increasing leading edge intensity (black arrow/inset) tracks the closing of the insulating gap as the rutile phase is reached, and a different aspect of the same physics yields to the decrease of the d||* feature at 530.6 eV (gray arrow/inset). The identity of the different peaks, together with a schematic representation of the corresponding density of states is given under the data of panel (b).
www.afm-journal.de www.advancedsciencenews.com
advantages are the presence of: i) the OK leading edge shift at ≈529 eV, ii) the directional d||* states in the monoclinic phase at
≈530.5 eV, and iii) the σ* states in the rutile phase at ≈531.5 eV which all show up clearly in the transmission XAS measure-ments, also without the need of a full suite of polarization-dependent measurements.
Summarizing, we report on the successful deposition of high-quality vanadium dioxide thin films on Si substrates and Si3N4 membranes using oxide nanosheets of TO and NWO as
a templating layer. Each nanosheet is a single crystal template, able to orient the growth direction of the VO2 film as a result
of lattice match of the oxygen frameworks of the two systems and compatibility of the 2D atomic structure between the nanosheets and the VO2 crystal planes. X-ray diffraction, SEM
imaging, EBSD, and ACOM-TEM data all agree on the film orientation, attesting to the epitaxial relationship such that TO nanosheets template (110)R growth and NWO templates (002)R
growth of VO2. It was also shown that even micrometer-scale,
local control over the VO2 orientation is achievable, when using
lithographically patterned nanosheet templates on a single, monolithic substrate.
The second main strand in the research presented deals with the MIT of the nanosheet templated VO2 films. Due to
strain effects along the c-axis of VO2 rutile phase, TMIT was
10 K higher (9 K lower) on TO NWO nanosheets, compared to bulk VO2 single crystal values. Truly bulk-sensitive soft X-ray
experiments carried out in transmission also underlined the excellent quality of the VO2 thin films, which displayed all
the hallmarks of the MIT known from films grown on bulk substrates. The significant soft X-ray transmission contrast—
for example, at the OK leading edge, or at the position of d||* feature—combined with the ability to work in
transmis-sion, means such nanosheet templated VO2 films are ideal
for advanced soft X-ray techniques such as lensless imaging of the MIT with spatial resolution of tens of nm. Such tech-niques involve holographic reconstruction of the real-space patterns formed during the MIT, and can be carried out on a modified version of the samples used for the studies reported here. Figure 8 shows a schematic for such an experiment. We have taken the first step toward such experiments and have recently successfully reconstructed first images during the MIT of VO2 using the holography with extended reference
by autocorrelation linear differential operator reconstruction technique.[36–38]
3. Conclusions
To conclude, we have shown that nanosheets can act as tem-plates for heteroepitaxial growth of high-quality VO2 thin films
on arbitrary substrates. The latter can be amorphous or have a very different crystallographic structure compared to the target material VO2 itself. This approach allows this important
test-case material for oxide-based devices and switching to be grown tailored for specific device applications and for funda-mental research. An example is given of how microstructured areas of differing VO2 orientation can be generated, and how
the nanosheet-templated growth approach can be used to enable soft X-ray holographic lensless imaging of the MIT in VO2.
Figure 7. V orbital occupancy across the MIT. a) Schematic of soft X-ray absorption experiments on TO-nanosheet-supported VO2 thin films grown on
20 nm thick silicon nitride TEM windows. Linearly polarized synchrotron radiation impinges in grazing incidence, as indicated (the transmitted beam is measured using a diode downstream of the sample, not shown). The c-axis of the (110)R-VO2 film is oriented differently in each of the nanosheet
domains, but is always in the plane of the film. Therefore, LH fixes E⊥crutile and LV polarization probes a mix of E⊥crutile and E||crutile.
Polarization-dependent measurements at the O-K edge above and below the transition for both b) grazing and c) normal incidence show the in-plane polarization of the unoccupied portion of the highly directionally aligned d|| states.
4. Experimental Section
Preparation of Nanosheet Films: Potassium carbonate K2CO3 (Fluka),
lithium carbonate Li2CO3 (Riedel-de Haen), titanium (IV) dioxide TiO2
(Sigma-Aldrich), niobium (V) oxide Nb2O5 (Alfa Aesar), and tungsten
(VI) oxide WO3 (Alfa Aesar) had a purity of 99.0% or higher and were
used as received. Nitric acid (HNO3; 65%, ACROS Organics) and
tetra-n-butylammonium hydroxide (TBAOH; 40% wt H2O, Alfa Aesar)
were used as received. Demineralized water was used throughout the experiments.
K0.8Ti1.73Li0.27O4 and LiNbWO6 were synthesized as reported in
the literature,[39,40] and the layered protonated titanate, H
1.07Ti1.73O4.
H2O, and protonated HNbWO6.xH2O were obtained by treating with
2 m HNO3 for 3 d and replacing new acid solution every day. The rapid
exfoliation with TBAOH, which has been reported for H1.07Ti1.73O4.
H2O[41] was also observed for HNbWO6.xH2O. Full-coverage TO and
NWO nanosheet films were generated on Si substrates and Si3N4 TEM
grips using the LB method.
The micrometer-scale patterned nanosheet template combining TO and NWO nanosheets was prepared as follows. A monolayer of TO nanosheets was deposited on a Si substrate. On top of this monolayer, hexamethyldisiloxane (Merck) was spin-coated at 3000 rpm for 30 s and then a thick layer of photoresist (OiR 907–12 from Olin Microelectronic Materials Inc.) was spin-coated at 3000 rpm for 30 s. After heating at 90 °C for 2 min, the sample was exposed to a Hg lamp with a wavelength of 365 nm for 10 s under a 20 µm spaced line grating mask (in a Karl Suss MA56 Mask Aligner). The photoresist was developed for 1 min (in OPD 4262 from Arch Chemicals) and baked at 110 °C for 2 min. Subsequently, a monolayer of NWO nanosheets was deposited on this line-patterned sample. Lastly, the sample was dipped and held
upside-down in acetone for 1 min for lift-off, in order to remove the NWO nanosheets on top of the photoresist. The sample was then rinsed with ethanol and dried in a N2 gas stream.
Pulsed Laser Deposition of VO2 Thin Films: PLD was carried out in a vacuum system equipped with a KrF excimer laser, with a wavelength of 248 nm (COMPEX from Coherent Inc.). The central part of the laser beam was selected with a mask and focused on a polycrystalline V2O5
target. The deposition conditions of the VO2 films were laser repetition
rate 4 Hz, energy density 1.3 J cm−2, spot size 1.8 mm2, oxygen pressure
7.5 mTorr, deposition temperature 520 °C, number of pulses 15 000, and substrate-target distance 50 mm. After deposition, the samples were cooled down to room temperature at a maximum rate of 5 °C min−1 at
the deposition pressure.
Analysis and Characterization: The surfaces of the nanosheet films
and of the VO2 on Si substrates were investigated using AFM (Bruker
Dimension ICON) operating in tapping mode and the data were processed using Gwyddion software (version 2.48).[42] The relative
coverage of nanosheets on substrates was determined at four different locations. The temperature-dependent crystal structure of VO2 thin
films was analyzed using XRD θ–2θ scans (PANalytical X’Pert Pro MRD) equipped with an Anton Paar DHS 1100 Domed Hot Stage. The temperature dependence of the resistance was measured using a Quantum Design Physical Properties Measurement System. A two-probe measurement was performed to extract the MIT behavior, hysteresis, and transition temperature. HR-SEM and EBSD were performed on a Merlin field emission microscope (Zeiss 1550) equipped with an angle-selective backscatter detector at room temperature. The ACOM-TEM mode was acquired on an FEI Tecnai G2 microscope (FEG, 200 kV), equipped with the ASTAR system from Nanomegas. Electron precession was applied to acquire quasi-kinematical data and to facilitate automated indexation. The precession angle used of 0.4° yielded an electron probe size of ≈1.5 nm. Interface mapping was achieved in post treatment with the orientation imaging microscopy (OIM) analysis software from Ametek EDAX company, including noise reduction and texture analysis. Misindexed or nonindexed points were corrected using a standard EBSD cleanup procedure. Furthermore, grains smaller than 5 pixels or with a low reliability (<10%) were removed from the analysis. The pole figures were calculated using the harmonic series expansion and were generated along the common directions (001, 110, etc.) with the Gaussian half-width set at 5°. The average orientation was taken from every identified grain.
Soft X-ray transmission experiments were carried out at the UE56-PGM1 beamline at the BESSY II synchrotron source located at the Helmholtz Centre HZB in Berlin. Linearly polarized X-rays (horizontal and vertical) were generated using the UE56 helical undulator, and the energy resolution of the beamline was set to 80 meV. The sample temperature was carefully controlled using a Janis cryostat, and the transmission of X-rays was monitored by comparing photon flux monitors before (refocusing mirror current) and after the sample (on a diode), and included correction for mirror/diode contamination by division with an “empty scan” in which no VO2 sample is held in the beam path. The soft
X-ray lensless imaging experiments were conducted using the COMET end-station at the SEXTANTS beamline of the SOLEIL synchrotron.[43]
Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.
Acknowledgements
P.T.P.L. and K.H. contributed equally to this work. The authors thank Mark A. Smithers for performing high-resolution scanning electron microscopy and electron backscattering diffraction. The authors also thank Dr. Nicolas Jaouen for his contribution to the soft X-ray imaging experiments. This work is part of the DESCO research program of the
Figure 8. Lensless imaging of the MIT of VO2. Schematic of a soft X-ray
holography experiment on nanosheet-supported VO2 thin films that
incorporate with a gold mask structure. The VO2 films can be grown using
PLD on TO nanosheets that are deposited on commercial, 200 nm thick silicon nitride TEM windows (e.g., Simpore). A gold mask is subsequently deposited and a sample window (diameter 2 µm) + reference slit are machined into the mask using a focused ion beam (FIB). The window reveals the VO2 film, whereas the slit goes right through the whole
struc-ture (including the VO2, the nanosheets, and silicon nitride, too).
Illumi-nation with coherent X-rays yields a far-field diffraction pattern, and using a differential filter and fast Fourier transform it is possible to reconstruct a real space image of the different phases of VO2 during the MIT. At
an X-ray free electron laser, sufficient intensity in a single ultrafast flash of X-rays would also allow a pump-probe version of this experiment, so enabling a stop-motion film to be built up of how VO2 switches on both
www.afm-journal.de www.advancedsciencenews.com
Foundation for Fundamental Research on Matter (FOM), which is part of the Netherlands Organisation for Scientific Research (NWO). P.T.P.L. acknowledges the NWO/CW ECHO grant ECHO.15.CM2.043. N.G. acknowledges funding from the Geconcentreerde Onderzoekacties (GOA) project “Solarpaint” of the University of Antwerp and the FLAG-ERA JTC 2017 project GRAPH-EYE. G.L. acknowledges financial support from the Flemish Research Fund (FWO) under project G.0365.15N. I.V. acknowledges support by the U.S. Department of Energy, Office of Science under Award Number 0000231415.
Conflict of Interest
The authors declare no conflict of interest.
Keywords
lensless imaging, nanosheets, transmission, vanadium dioxide, X-ray absorption
Received: January 2, 2019 Revised: July 26, 2019 Published online: October 31, 2019
[1] F. J. Morin, Phys. Rev. Lett. 1959, 3, 34.
[2] J. B. Goodenough, J. Solid State Chem. 1971, 3, 490.
[3] Z. Yang, C. Ko, S. Ramanathan, Annu. Rev. Mater. Res. 2011, 41, 337. [4] Y. Zhou, S. Ramanathan, Proc. IEEE 2015, 103, 1289.
[5] A. X. Gray, J. Jeong, N. P. Aetukuri, P. Granitzka, Z. Chen, R. Kukreja, D. Higley, T. Chase, A. H. Reid, H. Ohldag, M. A. Marcus, A. Scholl, A. T. Young, A. Doran, C. A. Jenkins, P. Shafer, E. Arenholz, M. G. Samant, S. S. P. Parkin, H. A. Dürr, Phys. Rev. Lett. 2016, 116, 116403.
[6] H.-T. Zhang, L. Zhang, D. Mukherjee, Y.-X. Zheng, R. C. Haislmaier, N. Alem, R. Engel-Herbert, Nat. Commun. 2015, 6, ncomms9475. [7] K. Martens, N. Aetukuri, J. Jeong, M. G. Samant, S. S. P. Parkin,
Appl. Phys. Lett. 2014, 104, 081918.
[8] J. Jian, A. Chen, W. Zhang, H. Wang, J. Appl. Phys. 2013, 114, 244301.
[9] J. Sakai, M. Zaghrioui, V. Ta Phuoc, S. Roger, C. Autret-Lambert, K. Okimura, J. Appl. Phys. 2013, 113, 123503.
[10] A. Gupta, R. Aggarwal, P. Gupta, T. Dutta, R. J. Narayan, J. Narayan,
Appl. Phys. Lett. 2009, 95, 111915.
[11] R. Molaei, M. R. Bayati, J. Narayan, J. Mater. Res. 2012, 27, 3103. [12] V. R. Morrison, R. P. Chatelain, K. L. Tiwari, A. Hendaoui,
A. Bruhács, M. Chaker, B. J. Siwick, Science 2014, 346, 445.
[13] M. Nakano, K. Shibuya, D. Okuyama, T. Hatano, S. Ono, M. Kawasaki, Y. Iwasa, Y. Tokura, Nature 2012, 487, 459.
[14] T. Driscoll, H.-T. Kim, B.-G. Chae, B.-J. Kim, Y.-W. Lee, N. M. Jokerst, S. Palit, D. R. Smith, M. D. Ventra, D. N. Basov, Science 2009, 325, 1518.
[15] E. Strelcov, Y. Lilach, A. Kolmakov, Nano Lett. 2009, 9, 2322. [16] H. Yoon, M. Choi, T.-W. Lim, H. Kwon, K. Ihm, J. K. Kim, S.-Y. Choi,
J. Son, Nat. Mater. 2016, 15, 1113.
[17] M. D. Nguyen, H. Yuan, E. P. Houwman, M. Dekkers, G. Koster, J. E. ten Elshof, G. Rijnders, ACS Appl. Mater. Interfaces 20168, 31120.
[18] M. Nijland, S. Thomas, M. A. Smithers, N. Banerjee, D. H. A. Blank, G. Rijnders, J. Xia, G. Koster, J. E. ten Elshof, Adv. Funct. Mater.
2015, 25, 5140.
[19] T. Shibata, H. Takano, Y. Ebina, D. S. Kim, T. C. Ozawa, K. Akatsuka, T. Ohnishi, K. Takada, T. Kogure, T. Sasaki, J. Mater. Chem. C 2013,
2, 441.
[20] T. Shibata, Y. Ebina, T. Ohnishi, K. Takada, T. Kogure, T. Sasaki,
Cryst. Growth Des. 2010, 10, 3787.
[21] M. Osada, T. Sasaki, J. Mater. Chem. 2009, 19, 2503. [22] J. Narayan, B. C. Larson, J. Appl. Phys. 2003, 93, 278.
[23] T. Shibata, K. Fukuda, Y. Ebina, T. Kogure, T. Sasaki, Adv. Mater.
2008, 20, 231.
[24] H. Yuan, M. Nguyen, T. Hammer, G. Koster, G. Rijnders, J. E. ten Elshof, ACS Appl. Mater. Interfaces 2015, 7, 27473.
[25] a) J. Li, J. Dho, J. Cryst. Growth 2014, 404, 84; b) Y. Muraoka, Z. Hiroi, Appl. Phys. Lett. 2002, 80, 583.
[26] a) D. Okuyama, K. Shibuya, R. Kumai, T. Suzuki, Y. Yamasaki, H. Nakao, Y. Murakami, M. Kawasaki, Y. Taguchi, Y. Tokura, T. Arima, Phys. Rev. B 2015, 91, 064101; b) K. Shibuya, M. Kawasaki, Y. Tokura, Appl. Phys. Lett. 201096, 022102.
[27] N. B. Aetukuri, A. X. Gray, M. Drouard, M. Cossale, L. Gao, A. H. Reid, R. Kukreja, H. Ohldag, C. A. Jenkins, E. Arenholz, K. P. Roche, H. A. Dürr, M. G. Samant, S. S. P. Parkin, Nat. Phys.
2013, 9, 661.
[28] L. L. Fan, S. Chen, Z. L. Luo, Q. H. Liu, Y. F. Wu, L. Song, D. X. Ji, P. Wang, W. S. Chu, C. Gao, C. W. Zou, Z. Y. Wu, Nano Lett. 2014,
14, 4036.
[29] A. Sohn, T. Kanki, K. Sakai, H. Tanaka, D.-W. Kim, Sci. Rep. 2015, 5, srep10417.
[30] C. Leroux, G. Nihoul, G. Van Tendeloo, Phys. Rev. B 1998, 57, 5111. [31] M. W. Haverkort, Z. Hu, A. Tanaka, W. Reichelt, S. V. Streltsov,
M. A. Korotin, V. I. Anisimov, H. H. Hsieh, H.-J. Lin, C. T. Chen, D. I. Khomskii, L. H. Tjeng, Phys. Rev. Lett. 2005, 95, 196404. [32] T. C. Koethe, Z. Hu, M. W. Haverkort, C. Schüßler-Langeheine,
F. Venturini, N. B. Brookes, O. Tjernberg, W. Reichelt, H. H. Hsieh, H.-J. Lin, C. T. Chen, L. H. Tjeng, Phys. Rev. Lett. 2006, 97, 116402. [33] F. M. F. de Groot, M. Grioni, J. C. Fuggle, J. Ghijsen, G. A. Sawatzky,
H. Petersen, Phys. Rev. B 1989, 40, 5715.
[34] M. Abbate, F. M. F. de Groot, J. C. Fuggle, Y. J. Ma, C. T. Chen, F. Sette, A. Fujimori, Y. Ueda, K. Kosuge, Phys. Rev. B 1991, 43, 7263. [35] N. F. Quackenbush, J. W. Tashman, J. A. Mundy, S. Sallis, H. Paik,
R. Misra, J. A. Moyer, J.-H. Guo, D. A. Fischer, J. C. Woicik, D. A. Muller, D. G. Schlom, L. F. J. Piper, Nano Lett. 2013, 13, 4857. [36] M. Guizar-Sicairos, J. R. Fienup, Opt. Express 2007, 15, 17592. [37] D. Zhu, M. Guizar-Sicairos, B. Wu, A. Scherz, Y. Acremann,
T. Tyliszczak, P. Fischer, N. Friedenberger, K. Ollefs, M. Farle, J. R. Fienup, J. Stöhr, Phys. Rev. Lett. 2010, 105, 043901.
[38] L. Vidas, C. M. Günther, T. A. Miller, B. Pfau, M. Schneider, E. Guehrs, R. E. Marvel, K. A. Hallman, R. F. Haglund Jr., S. Eisebitt, S. Wall, Nano Lett. 2018, 18, 3449.
[39] T. Sasaki, F. Kooli, M. Iida, Y. Michiue, S. Takenouchi, Y. Yajima, F. Izumi, B. C. Chakoumakos, M. Watanabe, Chem. Mater. 1998, 10, 4123.
[40] J. L. Fourquet, A. Le Bail, P. A. Gillet, Mater. Res. Bull. 1988, 23, 1163. [41] H. Yuan, D. Dubbink, R. Besselink, J. E. ten Elshof, Angew. Chem.,
Int. Ed. 2015, 54, 9239.
[42] D. Necˇas, P. Klapetek, Cent. Eur. J. Phys. 2012, 10, 181.
[43] H. Popescu, J. Perron, B. Pilette, R. Vacheresse, V. Pinty, R. Gaudemer, M. Sacchi, R. Delaunay, F. Fortuna, K. Medjoubi, K. Desjardins, J. Luning, N. Jaouen, J. Synchrotron Radiat. 2019, 26, 280.
